Re: reactance of 0805 and smaller SMT resistors

Eric Wheatley ([email protected])
Fri, 13 Jun 1997 16:45:20 -0700


I want to second your comments regarding vendor values of parasitics. I
have been concerned lately with the inductance of 0805 size chip capacitors
and have found some disagreement between values obtained by physical
parameter extraction and those reported by the vendor. I called the vendor
and found that the test procedure may not be optimal but has been maintained
to make the measurements consistent with reviously reported values for older
larger parts where the test procedure was appropriate. I agree that healthy
caution is advised when using catalog values of parasitic inductance
particularly when they are in the 0.1 - 1 nH range.

I would also like to point out that for these very tiny parts, the nature of
the contact (the amount and shape of the solder that connects the part to
the board) can be a significant part of the inductance. For parasitic
inductance, it is essential to specify or agree on exactly where the
terminals are that form the ends of the inductor. For example, consider a
part that is 80 x 50 x 50 mils with the terminals on the 50 x 50 mil ends of
the part. If one considers the entire 50 x 50 mil surfaces as the terminals
one will get the lowest possible inductance value. In reality, chips are
typically connected with a tapered blob of solder that sits on a trace. If
one uses the intersection of the solder with the trace as the terminal, one
gets a larger (and more realistic) inductance. I think similar
considerations apply to chip resistors.

I would also like to point out that inductance is a property of a chip
capacitor or resistor given a specific choice of terminals and a complete
description of the conductive materials in the device. This is called
partial inductance and it seems to cause a lot of confusion. No 'current
loop' is needed; just the acceptance of the fact that the current magically
appears at one terminal and disappears into the other. Of course chip
capacitors or resistors are never used in isolation but this does not negate
the possibility or value of assigning a meaningful value of inductance to an
isolated part. This is a useful parameter when, for example, selecting a
high speed bypass capacitor.

Eric Wheatley
Alterra Technology Co.

At 03:52 PM 6/13/97 -0700, you wrote:

>Charles Hill wrote:
>> Inductance is a property of a current loop, not a device. As such, the
current return path is as important as the signal traces. An 0805 device
mounted on the end of a coaxial cable has about 0.2nH of inductance; this
would be a minimum figure. This inductance is increased on a PCB with the
addition of vias and connections to the mounting pads. A smaller device
will have less inductance but that is not as important as how the device is
mounted, the trace lengths, and current return path.
>Sometimes it's easy to forget this when talking about the parasitics of
>devices. But having a device at the end of a piece of coax doesn't
>necessarily mean that a measurement of 0.2 nH is valid for all
>frequencies unless you account for the length and *parasitics* (losses,
>leakages, etc) of the coax, right? What about the added inductance
>of the path connecting the center conductor and "shield". It could be
>different if you are using RG58, RG8, or 0.085 semi-rigid due to the
>differences. Most of the component vendors I've seen that publish
>parasitics or frequency reposnse data also publish information regarding
>the fixturing and other components in the test setup. So generally
>when someone asks "what are the parasitics of X component", all the
>other factors you mention are hopefully understood and accounted for
>in both measurement and use of the device.
>For example there isn't any such thing as an unconditionally stable
>amplifier if you throw away the "no external feedback" assumption. No
>bothers to mention it every time, but sometimes people forget, so
>your reminder has value.
>BTW, there isn't anything magical about coax for this measurement.
>You could use coplanar transmission line, 300 ohm twinlead, or
>lamp cord depending on the frequency of measurement, availability
>of calibration standards, the desired accuracy and precision, etc, etc.
>Mark Randol, RF Measurements Engineer | Motorola SPS, Inc.
>(602)413-8052 Voice | M/S EL379
>(602)413-4150 FAX | 2100 E. Elliot Road
>[email protected] | Tempe, AZ 85284